We are working to calibrate a system that controls their high service pumps with VFDs at a water treatment plant based on a pressure sensor in the distribution system immediately downstream of the treatment plant. The target pressure is 50 psi. (HGL: 822.5). In order to calibrate the model we are modeling this location with a reservoir and flow control valve to specify flow rates that were measured during hydrant testing. However, due to the headloss through the valve, the simulated pressures within the system are lower than the observed pressures in the field.
The system has two elevated tanks and one standpipe which do not float off of the system pressure. The water mains within the system are fragile so they cannot operate at standard HGLs. Instead flow from the elevated tanks are regulated through PRVs and local pressure setpoints. All tanks and standpipes are filled via pumps with VFDs overnight. We have adjusted the elevated tank/standpipe elevations to mimic the pressure at which flows leave the elevated tank/standpipe.
Is there any way to model the treatment plant to specify pressure and flow and not experience an undesired pressure drop through the flow control valve?
Hello Megan,
Do you have FCV installed on field as well or you are adding that in the model only?
Generally there is a large headloss across the FCVs - this is the headloss necessary to balance energy across the model while satisfying all the constraints, including demands, boundary HGL and flow through the FCV. It could be that you have a large drop in HGL between an upstream and downstream boundary condition (is that the case ?) and with the desired FCV flow, the friction losses are less than this difference in boundary HGL and hence the rest of the headloss is seen through the FCV.
This can be interpreted as the FCV closing nearly all the way in order to restrict flow to the value you entered, while the boundary HGL conditions are still true. You can read more about it here:
Why do I see such a large headloss through my FCV, PSV or PRV?
A good way to model this could be to set a reservoir as a source with a PRV between it and the clearwell, with the PRV set at a HGL just below the top of the clearwell.
Here are few similar discussions, for your reference.
Modeling Treatment Plant connected to a Water Tank
Excessive headloss across FCV
Regards,
Sushma Choure
Bentley Technical Suppport
Megan Fox said:We are working to calibrate a system that controls their high service pumps with VFDs at a water treatment plant based on a pressure sensor in the distribution system immediately downstream of the treatment plant.
Instead of the FCV, you could use a pump set as a variable speed pump with a target pressure at a target node on the downstream side. See: Using Variable Speed Pumps in WaterGEMS and WaterCAD
If the resulting pump speed and flow is different from the measured conditions, this could mean that other model parameters like demands, roughness and valve status need to be adjusted to calibrate the model. This typically involves some investigation work first, and then potentially the use of WaterGEMS tools like Darwin Calibrator. See: Water Model Calibration Tips
Jesse DringoliTechnical Support Manager, OpenFlowsBentley Communities Site AdministratorBentley Systems, Inc.
Answer Verified By: Sushma Choure